One of the most significant public health challenges facing much of the industrialized world today involves the management of Alzheimer's Disease and associated dementias. Although advances in molecular biology hold promise for identifying drug targets that will slow the progression of several forms of dementia, our healthcare system remains limited in its capacity to effectively treat macro-scale deficits in cognition and language This is especially true with respect to progressive disorders of language and human communication, many of which reflect the loss of conceptual knowledge (i.e., semantic memory). To date, the development of theoretically-principled language interventions for dementia has been limited by two factors. First, neuroscience lacks a cohesive understanding of how regional brain damage impacts the organization and integrity of human semantic memory. Second, traditional neuropsychological rehabilitation approaches assume that learning is not possible in the context of progressive memory disorders. We seek to overcome these limitations through two specific aims. In Specific Aim 1, we will examine the potential for language maintenance over a two year period as we train patients with Alzheimer's Disease and Semantic Dementia on a carefully-crafted, personalized micro-lexicon consisting of approximately 100 words. Patients will train on this highly constrained vocabulary via a combination of repeated naming and semantic feature generation. This treatment approach is unique in that it protects a finite lexicon against loss rather than training forgotten concepts ad hoc as disease severity worsens. We will examine correlations between regional cerebral gray matter, neuropsychological performance, and language retention across time via pre/post structural neuroimaging and repeated administration of a comprehensive cognitive battery. Specific Aim 2 represents a complementary investigation of the nature of the mechanisms underlying naming impairment in Alzheimer's Disease and Semantic Dementia. We will investigate several aspects of visual search organization as participants name trained and untrained pictures. The high temporal resolution of eyetracking will yield essential insights into the relationships between visual search patterns, naming accuracy, and potential effect of treatment in these patient populations. We will also correlate these observed patterns of visual search with regional gray matter atrophy. We will contextualize these aims within the context of a broader model of semantic organization that correlates patterns of semantic impairment and learning in these clinical populations.